Asa $-$ Question #$3$: Does the correlation analysis indicate the presence of multicollinearity in the data?

Upon initial examination of the data, inconsistencies are observed, particularly with values such as $97,98,$ or $99,$ indicating respondents who did not provide an answer. There are two methods to address this issue: one involves replacing these values with the averages of the explanatory variable for the company, while the other entails removing the entire row of values. I $($Dr$.$ Parmer$)$ recommend the latter approach.

To refine our correlation analysis, certain variables are excluded based on domain knowledge. Variables such as 'Likelihood to Repurchase,$\text{'}$ 'Likelihood to Recommend,$\text{'}$ and 'Recommended E$-$Commerce Site to Family and Friends in the Last $3$ Months' are omitted as they are seen as direct consequences of high customer satisfaction rather than contributors to its explanation.

Additionally, 'Overall Experiences SATISFACTION with $($INSERT NAME$)\text{'}$ is excluded due to its broad construct lacking actionable insights. Categorical variables like 'Method Used Most Frequently to Shop at $($INSERT NAME$)\text{'}$ and 'Method of Payment Do You Prefer Most for Shopping Online $($INSERT NAME$)\text{'}$ are also omitted for accuracy, given that correlation analysis typically involves continuous variables.

Furthermore, 'Overall Service Quality' is not considered due to its potential limited actionability, and there is an argument for its exclusion from the analysis.

In the final step, it is advised to check for multi$-$collinearity by running a correlation analysis on the cleaned Amazon data set. The correlation matrix, including labels in the first row, should be included in the deliverable. If the correlation values between explanatory variables are not greater than $0\mathrm{.}7,$ this would rule out the possibility of multi$-$collinearity.